Cemented hydraulic backfill (CHB) is widely used in underground mine backfilling, especially when regional stability is required. One of the critical backfill properties is uniaxial compressive strength (UCS), the maximum axial compressive stress that the sample can withstand before failing. CHB is expected to have a certain UCS in a specific time so that the adjacent stope can be mined out according to the production plan. To reach the desired strength in a specific time, there exit parameters which are must be well adjusted. To increase the strength of CHB, either the cement dosage could be increased or a longer curing time could be allowed using less cement and more tailings. However,increasing cement content significantly increases the operational mining cost. If there is enough curing time for the planned production then less cement and/or more tailings can be added to get the desired strength at a reduced cost. This paper investigates the applicability of artificial intelligence (AI) algorithms to optimise key parameters of CHB design so that the desired strength would be reached in a specific time. Genetic programming (GP) is used to generate models relating the UCS factor to CHB’s key parameters using an experimental database. The generated GP models are then used by a particle swarm optimisation (PSO) algorithm in order to determine the amounts of CHB’s parameters which can satisfy the specified UCS conditions in a planned time. Some examples are presented to emphasize the benefits of the optimization of CHB mixture design. Using the presented approach, it is possible to optimize CHB design parameters by considering mine production plan, requiring certain UCS at specific ages, and to reduce the cost.
|Springer Series in Geomechanics and Geoengineering
|28th International Symposium on Mine Planning and Equipment Selection
|2/12/19 → 4/12/19