Discrete Element Modeling (DEM) is not a plug and play tool

Historically, chutes were designed based on experience and trial-and-error methods. These designs were checked, with no guarantee to work, by building scaled-down models that are not only time and cost prohibitive but also difficult to scale. The major difficulty arises due to the fact that unlike fluid flows, there are no scaling laws for bulk solids.  In particular, cohesion and frictional forces in a large-scale operations are different in comparison with a small-scale model. Using a dimensionless quantity such as a Froude number alone (ratio of inertia to gravity force) is flawed due to the mere fact that cohesion and friction forces are not included in this number.  As such, this methodology does not include comprehensive dynamic scaling and most importantly, is highly risky for the time and cost put into it.  More often, these approaches have a strong element of trial and error, and benefit from a great deal of on the job experience, but are limited in their application and usefulness.

DEM technology has been used as a research tool for many applications. However, just recently it is gaining popularity as a major virtual design tool for large chutes due to its ease of use and cost efficiency in comparison with classical approaches, including the scaled-down model. Although a very useful and powerful tool it has challenges and limitations.  In particular, just as other computational predictive tools such as CFD or FEA, it needs to be calibrated; and garbage in to the model is garbage out of it. It is not a plug and play tool and the input and selected models need to be well thought through, and its impact on the bulk behavior needs to be understood. The contact and cohesion force laws used will vary the inputs significantly. Having experience with different bulk solids behavior and an understanding of the various contact laws and associated inputs is imperative for the DEM model to predict the flow in the chute.

6 Responses to Discrete Element Modeling (DEM) is not a plug and play tool

  1. Richard says:

    An interesting article, I agree with the points made. I would be very interested to hear more about the cohesion force laws you are applying in your DEM models and your approach to calibrating cohesive bulk material properties.

  2. Steve Davis says:

    I agree with your comment, and have been recommending this for some time to anyone who will listen.
    Peter Wypych and Andrew Griema at BEMA have been calibrating for some time with EDEM and Rio Tinto. TUNRA are also able to calibrate bulk materials for DEM.
    Are Jenike and Johanson also able to offer this service?

  3. D V Behere says:

    What impact chute liners and chute material will have on sizing and material flow. Standard liner material used are TISCRAL, SS and UHMP, HARDOX and selection made depending on material properties.

    • Rahul Bharadwaj says:

      The chute linear and wall surface material play an important role in the efficient operation of the chute. Selecting a good wall material will effect the flow pattern and also the wear on the surface. We first test the materials on the surface of the material using our shear tester and then use this as one of our inputs in our DEM models. In addition, we use our wear test results couple with our DEM models, to predict the surface life and wear patterns on the desired chute wall material.

  4. Richard Elliott says:

    Steve, It is my experience as a user of DEM software that the simulation of a cohesive material is still a significant challenge. I am familiar with the materials testing services available at Tunra and BMEA and have used both in the past, getting the material of interest tested is definitely an essential requirement.

    The difficulty I have found is the normal force laws within many commercial DEM codes often lack the ability to have an adhesive force that is a function of consolidation history or time for example. I know developers and academics are actively pursing this area but I have not seen much published which shows DEM of a transfer or bin that handles a cohesive material where the DEM was able to realistically predict the flow problems observed. I was curious to know more about the approach taken by J&J?

  5. Rahul Bharadwaj says:

    Hi Steve/Richard,
    It is indeed a challenge to have a calibrated and validated DEM model. We have lots of experience at J&J regarding the way materials flow and behave and do calibrate all the inputs to our model. As I was saying, it is imperative to understand what contact laws are used for interactions and also how sensitive each parameter is to the bulk behavior of the material. For example, we have our proprietary cohesion law that we use and the inputs to the model are obtained from small scale calibrated experiments in our lab. We do offer these calibration services to our clients and have been working with large mining companies as well. In addition to mining, we also perform DEM model development for other processes and industries and we do use our extensive lab capabilities to get inputs to our models.
    Would be glad to provide references and to discuss more if you have any specific application.