Open-pit mine planning with operational constraints

Open-pit mines must be designed to develop the Earth’s natural resources in the most responsible, sustainable, and economic way. Traditional mine planning optimization methods do not consider operational constraints; such as minimum mining width or minimum pushback width constraints, and often do not generate realistic, actionable designs. This dissertation develops techniques to incorporate operational constraints into open-pit mine planning which allows for engineers to more accurately convert mineral resources into mineral reserves and better evaluate the economic viability of open-pit mining projects. A major practical challenge is that the resulting mathematical models are very large, with potentially hundreds of millions of variables and constraints. Addressing this challenge and delivering tools which are usable on real-world 3D datasets requires a theoretically motivated and computationally grounded approach. The first contribution of this dissertation is an efficient implementation of the pseudoflow algorithm for the well known ultimate pit problem. Modest theoretical improvements and practical computational improvements combine to create a fast and efficient open source ultimate pit optimizer, called “MineFlow,” which is more performant than all evaluated commercial alternatives. A model with sixteen million blocks which takes over three minutes to solve with a commercial ultimate pit optimizer is solved in nine seconds with this implementation. The second contribution is a formulation and methodology for the ultimate pit problem with minimum mining width constraints. These operational constraints restrict the shape of the ultimate pit in order to provide suitably large operating areas which can accommodate the large machinery used in open-pit mining. This problem is shown to be NP-complete and several optimization approaches are developed in order to compute high quality results for large block models in a reasonable amount of time. The two most effective approaches use Lagrangian relaxation and the Bienstock-Zuckerberg algorithm which are modified for this problem. Moreover, the formulation is extended to open pit direct block scheduling problems with operational constraints and solved using a newly developed method based on the Bienstock-Zuckerberg algorithm. This approach is applicable to large, realistic, open-pit planning problems that span multiple time periods and multiple possible destinations for each block.

Name thesis:degree_name

Doctor of Philosophy (Ph.D.)

Level thesis:degree_level

Doctoral

Discipline thesis:degree_discipline

Mining Engineering

Grantor dc:publisher

Colorado School of Mines. Arthur Lakes Library

Year dc:date.issued

2023

Author dc:creator

• Deutsch, Matthew Vernon

Advisors dc:contributor.advisor

• Dagdelen, Kadri
• Johnson, Thys B.

Committee members dc:contributor.committeemember

• Newman, Alexandra M.
• Mehta, Dinesh P.
• Kaunda, Rennie

Statement dc:rights

• Copyright of the original work is retained by the author.

Language dc:language.iso

eng, English

Identifier

T 9655

OAI identifier oai:identifier

oai:repository.mines.edu:11124/178655