Clients
SELCO Is Developing Optimization Tools for ESAN
Mine Scheduling Optimization [2022]
ESAN Planning Teams at Balya Mine, monitor their operations in real-time and need an scheduling / optimization tool that can help them to reschedule their production activities, while meeting all sort of conflicting constraints. The planning of underground mining operations require a set of activities that should be carried out meticulously. While minimizing inefficiencies, the production schedule should be able utilize not only the teams, but also the very expensive underground machines and trucks required to carry material in a single truck width road network. This is an extremely hard problem, which requires other innovative techniques as neither optimization nor any other scheduling algortihm can successfully solve to optimality.
ESAN chose to work with SELCO to solve this unique problem and the end result was quite innovative. Following the analysis of the existing processes, the SELCO Analytics team designed the analytical model that would solve the assignment and scheduling problem in the fastest and optimum way. The complex/multistage optimization model was developed on an OPL code running on IBM CPLEX engine.
After the seamless integration of the optimization model to their operational system, ESAN teams can now create, in real-time, the optimum scheduling for their operations at the Balya mine. The value created by the optimization solution are threefold:
- 15 % reduction in process cycles
- 100 % adherence to compulsory tasks
- Elimination of idle times at mning machines
OPTIMIX Project [2020]
ESAN
Established in 1978 to produce high quality raw materials for the ceramics sector, ESAN became one of Turkeys foremost industrial mineral and metallic mineral producers. The plants for clay enrichment, feldspar flotation, quartz, bentonite, kaolin, magnesium metal, cat litter transform the mineral resources to high quality ingredients in many sectors. To gain momentum for analytics, blending optimization was chosen as a compelling business problem, where optimization techniques were expected to produce faster and better results, the Bozüyük plant being the first location to do the pilot study.
Timuçin BÜLBÜL, Business Development Director, ESAN: "With the help of Selco Consulting, IBM CPLEX enabled our blending process to have an optimal answer from a space of countless solutions. This methodology showed us that optimization methodology can be very useful in many sectors, where very complex and unsolved problems of dozens of parameters can be solved within seconds, more accurate than human-being could ever do."
Blending planning process
The goal of blending is to achieve optimum mix of particle sizes and balance the variation of quality parameters. The actual crushing and mixing of the raw materials is preceded by a planning activity, when the planning team spends time to find an optimum mix recipe for the blending batch, trying to come up with the lowest cost of raw materials, while meeting the quality parameters and also a recipe that has a low filter press cycle time. The planning process has several dimensions which need to be addressed simultaneously. Raw material dynamics include batch sizes, limits on recipe contents, and inventory levels. Production capacity requirements favor those recipe contents that require less process times, as cheaper but longer process times may impose a bottleneck on plant throughput.
Tool
A sophisticated solution was developed to provide insight to creating optimum blending recipes; the result is an instantaneous opportunity to view optimum material costs and product quality reflections. The tool was developed on CPLEX and was supported with user friendly interfaces, allowing the users to move from managing the data to calculating the optimum blend which respects all the business requirements. It also provided a unique sensitivity analysis approach, whereby the planners can very elegantly identify how the cost and quality indicators tend to behave with different solution sets and relaxed physical/chemical constraints.
Results
The analytical model with its carefully balanced performance tuning is able to map an entire cross section of the feasible solution space, enabling the quants to better understand the dynamics of how cost improvements were made. The solution map was shown to produce a set of optimal solutions having zero-gap tolerance, despite the non-linear and quadratic nature of the problem. Aside from reduced costs, the tool reduced the new product development cycle and allowed the planners to dedicate their times to value added and innovative tasks.
For more information...