INTEGRATED SMART MINING APPLICATIONS (ISMA)

ISMA is a dynamic algorithm design by Dr. Yassiah Bissiri, a University of British Columbia (UBC) graduate and founder of INAREC, for mining /natural resources projects

  • The concept of integrating all mining modules for the purpose of optimization became a hot topic among mine planners in the early 2000. However to this date, an effective integration of all modules at once which represent the best platform for the true optimization of an entire mine is yet to be implemented
  • ISMA is a blackbox that not only integrates the modules but it also uses a novel approach to bring best solutions in a dynamic manner (Best solutions are proposed at the time decision making)
  • Mining is not manufacturing

INTEGRATION OF MINING MODULES

SMART MINING

  • Several methods of optimizing processes exist today.
    • Hardcore math based optimization:
      • Consist of building a mathematical model (predictive, uncertain environments, divide and conquer methods….)
  • They can prove to be very complicated.
    • Example of the divide and conquer method:
    • One may divide and optimize each modules. However one may lose significant parameters while regrouping these resulting modules and might end up solving everything but the initial problem
  • Rule Base or Heuristic approach
    • Rules are set by experienced experts and a model is built to encompass the procedure
    • it could be short sighted by the nature of the operational environment (very changing) and bias the all process
    • Intelligent systems based
      • By far the best approach given the availability of technology and bandwidth today that makes it easier to implement
  • However it is important to define with precision what is that we are trying to optimize. Doing so is the most important step in the process of finding good solutions at the time a decision is needed
    • Capturing the complexity of a process solves 50% of the problem and this is done by identifying significant parameters that makes the project
    • Interfaces between modules
      • Interaction between parameters of significance of different modules needs to be model and translated into a computer algorithm
    • Approach works for every mining methods

EXEMPLE OF INTERACTION BETWEENS TRUCKS AND SHOVELS MAINTENANCE MODULES WITHIN ISMA: CASE OF EARTH MOVEMENT OPERATIONS IN MINING

  • The proposed approach will yield solutions that are:
    • Flexible: one can add and retract parameters based on the evolution of the working environments
    • Self healing: the system will always try to survive should a catastrophe, such as losing several pieces equipment/machineries to unusual massive breakdowns.
    • At the time of decision making, the solution proposed by the system will be the best of the moment until next decision time.
  • The method uses competitions among different participating Agents in the operational environments for resources and a resource is awarded to the agent that satisfy the objectives (Static or dynamic) at the time of decision
    • Don’t we hear all the time clean competition is good for excellency?
    • In operational environment such as massive earth moving job, operational objectives changes all the time and pseudo uncertainty-based or predicitive-based optimization often fail to capture that complexity and may end up making the solving process complicated
  • Resources and Competitors interact among each other
  • Simple mechanisms through cooperation
    • Competitors cooperate in order to satisfy current objectives
    • Survival of the system, which is linked to the objectives, is key
      • Meaning that the system has the ability to “auto-heal itself”
      • Under pressure from breakdowns and unforeseen events, System will always re-arrange to find the best solutions given the situation
    • Unlike other systems, functions are simple to generate as capturing the complexity of the operational environment is now a function of the “intelligence” brought to it.
    • Difference between maintenance module and shop. Maintenance module assists with predictive maintenance whereas the shop requires the physical removal of the resource.
  • Maintenance module (mm) and shop (mm) will compete for shovels and trucks (resources) by generating demand values. 
  • Shovels and trucks will react to each demand by generating a likelihood to satisfy function.
    • A response function value is calculated from the competitors and the resource is then allocated to the competitor with the highest response value.