Personal tools
You are here: Home News Print Editions July-August 2010 Q&A with John Favier, chief executive, DEM Solutions

Q&A with John Favier, chief executive, DEM Solutions

by Australian Bulk Handling Review last modified Oct 15, 2010 03:13 PM
— filed under:

Discrete element modelling (DEM) software is fast finding favour for the design of bulk handling equipment and simulation of bulk flows. But significant hurdles remain, particularly in terms of calibration and validation. ABHR editor Charles Macdonald spoke to John Favier of leading software developer, DEM Solutions.

Q&A with John Favier, chief executive, DEM Solutions

John Favier, chief executive, DEM Solutions.

Q - Why should the bulk handling sector invest in DEM simulation?

A - The reliability and durability of bulk handling equipment is critical and producers are looking to increase availability and in-service life to ensure that they get the best return from capital investment and can improve their ability to respond quickly to market opportunity.

Re-work of a single conveyor transfer point to remedy problems with blockage or spillage can cost suppliers hundreds of thousand of dollars. The cost of delays and downtime to a mining operation are often even greater. Feed to crushers and sieves, operation of reclaimers and loading devices are other sources of potential problems.

Sub-optimal production rates as a result of problems associated with increasing throughput or changes to the bulk material characteristics can also be hugely expensive. Such problems are not new to the industry and the mining sector in particular is looking closely at using DEM simulation as an alternative to the traditional approach to design of bulk handling equipment based on empirical design rules and physical prototyping using scale models.

As with other industrial application of engineering simulation tools, DEM simulation does not replace the engineer’s equipment and process design know-how. When carried out properly, it allows “what-if” scenarios to be explored up-front and increases confidence in the performance of the equipment in the field. The facility to explore the design space reduces risk both to the equipment supplier and the end-user. More advanced applications are now using DEM to estimate equipment wear and fatigue; information which is valuable in estimating in-service life and determining maintenance schedules.

Despite these benefits, as implied in your next question, some industry participants remain cautious about whether DEM can produce results of sufficient quality to be useful for engineering design. If DEM were entirely new to the industry then this could be explained as typical of the adoption cycle of any new, unproven technology. I have outlined below some of the reasons why calibration is seen as a block to acceptance of DEM. It is a legitimate concern that has not been properly addressed by many DEM practitioners. As more information and guidance on calibration enters the public domain, and more examples of the equipment designed using DEM are commissioned and are seen to perform successfully, these concerns should reduce.


jul-aug-edm1

EDEM simulation identifies the combination of chute design and operating conditions that will cause blockage of a transfer point handling high moisture coal.


Q - Industry participants tell me that to boost the acceptance of DEM, further work needs to be done in terms of calibration procedures. Can you address this issue? And also the issue of the need for large-scale validations?

A - We are seeing confidence in DEM growing among leading companies in the industry including OEMs, EPCs and mining houses within our customer base. They are all using calibrated models in their designs.

While it is the case that calibration of DEM models is still not very well established, the problem in the past has often been lack of calibration rather than lack of available methods for calibration. The reason for this seems to have been a combination of the limited capability of the earlier generation of DEM software to adequately characterize the bulk material in terms of particle size and shape and physics, and the suggestion by some providers of services that DEM simulation “expertise” can substitute for calibration.

DEM Solutions has pioneered the concept of the “DEM material model” which is the combination of the particle size and shape distribution and contact model that determine the bulk behavior of the material in the simulation. It is this material model that must be calibrated and methods for efficiently determining the optimum parameter set are needed as well as the physical calibration tests.

Much of this work has not been published due to the market advantage that our customers are seeing from having an advanced DEM simulation capability incorporating more accurate, calibrated solutions than previously available in the market. However some of our academic collaborators are beginning to publish key papers in this area showing that with a more sophisticated approach it is possible to get very good results for flow simulation of bulk materials such as coal and mineral ores. Importantly, the physical tests being used are relatively simple and do not require expensive measurement equipment.

I see parallels with the early days of finite element simulation when the material models for use in simulation of the structural behaviour of materials such as steel, concrete and so on were not generally available and procedures were developed which combined physical testing with model calibration methods that simulated the same tests. The large variability of bulk materials means that it will be a long time before calibrated material models are available off-the-shelf but the conceptual approach is useful for DEM and is proving to work well in the field.

One of the powerful aspects of DEM simulation is its scalability once the particle-level model has been calibrated. Calibration procedures are typically carried out at bench-scale but, provided they are properly chosen, the calibrated model will reproduce bulk flow at process scale.

The question of validating at large scale is an interesting one when looked at from the perspective of the performance and in-service life of equipment designed utilizing DEM simulation and equipment designed using more traditional methods. We observe that equipment suppliers do not as a rule carry out tests or make measurements to confirm that their installed equipment is performing to specification in the field regardless of which design rules and tools are used. Only if equipment is malfunctioning and rework is required are the underlying causes of the poor performance examined. The difficulty and cost of making such measurements in the field is of course a constraint on validating designs at large scale whether DEM simulation has been used or not, but it is certainly an area which could benefit from attention by instrumentation suppliers.


jul-aug-edm2

Using EDEM simulation to test the performance of a prototype design of a rotating charge distribution chute in a blast furnace. The simulations show the material flow and particle size segregation of the charge layers of iron ore and coke.


Q - More realistic particle shape representation is another call from industry. What’s happening in this area?

A - Particle shape representation is an aspect of DEM simulation that is often misunderstood. Intuitively it would appear that in order to simulate ROM coal for example, which may comprise a wide range of particle shapes, often quite angular, it is necessary to use the same particle shapes in the DEM model. However, while it is possible to represent bulk materials using complex polyhedral shapes it is not necessary to do so in order to predict the bulk flow of these materials with a degree of approximation to reality sufficient to provide useful engineering data.

The bulk behaviour of a bulk material is dependent on a range of factors besides the particle shape. When taken together, as defined within the DEM material model, then the flow behaviour in the simulation will be sensitive to particular parameters depending on the nature of the material. The dominant shape metric is the aspect ratio and, roughly speaking, as this becomes more extreme then the DEM model particle shape needs to more closely match the actual particle shape in order for the bulk behaviour in the model to match the physical system.

In fact, for the majority of bulk materials handling by the mining industry, certainly for bulk flow such as a conveyor transfer operation, it is possible to get very good results by using material models employing particle shapes based on spherical particle elements. Models confined only to spheres will generally not produce sufficiently accurate results to be used for detailed design work but can be useful for screening of designs at an early stage when changes to equipment geometry in design prototypes are very large.

Simply put, the more complex the particle shape, the greater the compute effort required to simulate that shape. As the price-performance ratio of computer hardware improves then we will see more complex particle shapes being introduced. However the demand for such shapes should be driven by the desired quality of the engineering data produced rather than a priori assumptions that unless the particle shape is more “realistic” then DEM simulation will not be successful.


jul-aug-edm3

Location and degree of abrasive wear of a charge distribution chute predicted by EDEM simulation. The analysis is used to determined the required wear protection and estimate the maintenance schedule.


Q - Can you briefly explain the advances in DEM and its application to the bulk handling sectors since its first appearance? What have been the major advances over the last 10 years?

A - DEM was first introduced in the 1970s as a computational technique for application in geomechanics research. It was 2D and confined to very small numbers of particles. With the advent of workstation computing in the early 1990s DEM started to be used in research into bulk handling applications such as hopper discharge and conveyor transfer and more widely for fundamental research in particle technology.

The use of DEM in academic research grew exponentially throughout the 1990s and a considerable body of knowledge was developed relating to the physics of bulk materials at the particle-scale. Most DEM software up until the end of the 1990s, with some notable exceptions amongst research codes, were very limited in their ability to handle anything but the most simple equipment geometry in 3D. Most commercial DEM codes were 2D although it is generally accepted that it is necessary to simulate bulk solid flow in 3D to obtain satisfactory results.

The other major constraint for application of DEM in the bulk handling sector until relatively recently was the scale of models and compute time when software was run on the type of desktop computers employed by design engineers in the industry. The advent of high performance computing on multicore desktop computers has changed DEM from being largely a research tool to being a feasible technology for use in engineering design.

The other major advance over the last 10 years has been the development of DEM software with a user interface specifically developed to make the simulation technology accessible to design and process engineers.

Q - To replace prototyping and practical experimentation even more broadly, what advances in DEM are needed over the next 10 years?

A - No one nowadays would consider designing aircraft, cars and trucks, chemical process plant or an oil production pipeline to name but a few, without using engineering simulation technology. DEM simulation is a core technology for application to bulk materials handling and processing. Its application is still at an early stage with a relatively small number of industry leaders getting the benefits. This is because it requires know-how as well as quality software to profitably deploy DEM. However with good support it is already possible for companies to bring DEM in-house – this is particularly important for OEMs and EPCs where DEM simulation generates valuable IP they need to keep in-house. Producers are beginning to demand a greater level of predictability of equipment and system performance from their suppliers. Improving the availability of plant and estimation of in-service life is of great value to producers and the next 10 years will see DEM becoming an essential tool deployed to support delivery of superior products.

DEM is beginning to be deployed in a similar manner to more established computer-aided engineering technologies such as finite element analysis and computational fluid dynamics. This type of software is part of a virtual prototyping toolkit that reduces dependency but doesn’t necessarily remove the need for some physical prototyping. The bulk handling industry has traditionally relied on physical prototyping and empirically- based design rules to develop equipment and processes. This has constrained innovation and limited exploration of more efficient design due to the high cost and time required for physical experimentation and prototyping.

Q - What’s your background and history with DEM?

A - I have been involved in numerical simulation my entire professional career. I became interested in DEM in the mid-1990s as part of R&D of equipment for handling and primary processing of agricultural materials such as grains and soils. To be able to simulate bulk handling of these materials I needed to improve on the state-of-the-art at that time by developing the capability to model non-spherical particle shapes and then to incorporate real equipment geometry into DEM simulations.

The new capability was quite generic with application to a wide range of industry problems and attracted interest from a number of companies looking for new solutions to particlerelated problems. This industry demand led eventually to my founding DEM Solutions in 2002 to further develop the prototype software for use by industry. My vision was to create a new form of user interface for DEM simulation that provided the components of the workflow needed to create, solve a model and analyze the results as well as providing a platform for implementing as wide a range of simulation solutions as possible.

Q - Tell me about the importance of seamless integration with CAD and developments in this area?

A - Enabling a user of DEM to easily incorporate CAD models of equipment into their simulation model is essential for industrial application of DEM. We have taken the approach of providing as much flexibility in the range of CAD model formats that can be imported into our software. This provides good CAD compatibility while not going as far as full integration. The latter is possible and has the benefit of enabling fast update of changes to the CAD model within the DEM simulation. The disadvantage is that it is usually CAD-product specific, which forces the customer to employ a particular CAD package in order to use the DEM software. It puts the focus on the CAD, which may be at the expense of the quality of the DEM. An emphasis on CAD-integration can make sense for low-fidelity DEM models used for early screening of design prototypes where the user is not expert in DEM and the capability of the software is limited.

Q - What are the hardware requirements for DEM programmes? And what are the computation times for complex problems?

A - Commercial DEM software typically runs on desktop computers with compute performance related to the ability to utilize parallel processing on multiple processors. DEM Solutions software is parallelized to run on multi-core computers and can comfortably handle bulk materials handling applications such as a conveyor transfer point with mass flow rates of several thousand tonnes per hour of ore or coal. This level of compute power is required to implement the material models (particle size, shape and number) needed to produce results of sufficient quality for engineering design.

Q - Tell me about the importance of seamless integration with CAD and developments in this area?

A - Enabling a user of DEM to easily incorporate CAD models of equipment into their simulation model is essential for industrial application of DEM. We have taken the approach of providing as much flexibility in the range of CAD model formats that can be imported into our software. This provides good CAD compatibility while not going as far as full integration. The latter is possible and has the benefit of enabling fast update of changes to the CAD model within the DEM simulation. The disadvantage is that it is usually CAD-product specific, which forces the customer to employ a particular CAD package in order to use the DEM software. It puts the focus on the CAD, which may be at the expense of the quality of the DEM. An emphasis on CAD-integration can make sense for low-fidelity DEM models used for early screening of design prototypes where the user is not expert in DEM and the capability of the software is limited.

Q - What are the hardware requirements for DEM programmes? And what are the computation times for complex problems?

A - Commercial DEM software typically runs on desktop computers with compute performance related to the ability to utilize parallel processing on multiple processors. DEM Solutions software is parallelized to run on multi-core computers and can comfortably handle bulk materials handling applications such as a conveyor transfer point with mass flow rates of several thousand tonnes per hour of ore or coal. This level of compute power is required to implement the material models (particle size, shape and number) needed to produce results of sufficient quality for engineering design.

Compute times are from hours to days depending on the complexity of the physics and rate of throughput. Lighterweight DEM codes use cruder models to represent the bulk material, often sphere-only particle shapes with compute times in days to weeks if they attempt to run larger flow rates. There are DEM research codes that run on large-scale parallel computing systems and can handle larger scale problems but they are usually optimized to particular applications and are only useable by those that wrote the software

Q - What’s the cutting edge of DEM nowadays in terms of solving bulk handling flow problems, and designing things like conveyors, chutes and silos?

A - Particle-structure interaction (PSI) is one of the new areas of focus in DEM simulation of bulk materials handling. We are seeing a lot of interest in using DEM to predict the rate of wear of equipment such as transfer chutes, rock boxes and to estimate the form and level of wear protection required to ensure a given in-service life. DEM can provide the magnitude and location of impact and abrasive forces but the challenge is to determine the correct wear functions for given bulk material and equipment lining material interaction. Poor design of wear protection can sometimes result in components wearing out much more quickly than expected with consequent high cost of replacement.

Another PSI-type capability we have been working with customers to develop is integration of DEM with structural analysis to provide the static and dynamic loading conditions for improved design of structures supporting conveyor transfer points and silos for example. DEM can provide key design data as input to dynamic analysis of loading of crushers and sieves. Asymmetric loading from feed conveyors and transfer chutes is often a cause of excessive stresses and fatigue-related equipment failure.

Prediction of dust generation and design of dust mitigation and control systems are another area at the cutting edge of DEM applications. The prediction of dust is very challenging as it requires an understanding of how to model the interaction between dust loading, air entrainment and bulk flow characteristics. One aspect of this type of problem we have worked with a customer to develop is using coupled DEM-CFD simulation in the design of dust extraction systems. One of our customers has also used this capability to design a system for removal of pollutant gases generated during loading of bulk material containing volatiles.

Q - Looking into your crystal ball, what might DEM be able to do in 10 years’ time?

A - Technical advances that will drive the use of DEM will be increases in compute performance to shorten the turnaround time for design iterations based on DEM simulation. This increased performance will come from utilization of new generations of multi-processor hardware technologies that are currently at an early-stage of commercial deployment. As these become more standard and available to engineering design then DEM software will be able to take advantage of the compute power to increase the scale and complexity of processes modeled as well as reduce compute times.

Methods for calibration of DEM material models will become standardized including physical tests and related model optimization procedures. Tools and services to support material model calibration will be more widely available.

DEM will continue to be better integrated with CAD and other CAE simulation technologies. There are already a lot of know-how and modelling techniques developed in research. The driver for such development of commercial capability will be the engineering need and value.


jul-aug-edm4

EDEM simulation of conveyor transfer showing locations of the wear of the hood and spoon chutes.


Q - There is already some collaboration between the software companies and universities. But could there be closer involvement and what might be the possible benefits of this?

A - Universities have an important role to play in applied research and in training of engineers in the application of DEM to industrial problems. Developing methodologies and standards for application of DEM needs good collaboration between academic research, software companies and commercial users of DEM software. I noted earlier that DEM Solutions is active in this type of collaboration as we see it producing a good return for industry. There are now over 150 academic research groups worldwide using our simulation
technology, many of whom work in the area of bulk materials handling. These groups usually already have experience in
measurement and characterization of bulk solids and some of the leading groups are working to further develop and benchmark
DEM calibration methods. Their expertise in calibration provides an opportunity to provide a service to industry in carrying out the necessary tests and training in how to carry out calibration procedures.

We recently released a teaching version of our academic software for use by universities in teaching the fundamentals of bulk materials flow and the basics of application of DEM simulation to solving bulk materials flow problems. The benefits to industry will be engineering graduates with a better knowledge of DEM simulation and how it can be deployed in solution of industry problems.

Contact: John.Favier@dem-solutions.com





Document Actions
  • © 2013 The Australian Bulk Handling Review