A decision support system to optimise the available resources at Kimberley Mines

Du Toit, Jeremias Cornelius (2007-03)

Thesis (MBA)--Stellenbosch University, 2007.

Thesis

ENGLISH ABSTRACT: Diamond mining started in Kimberley in the early 1870s following the discovery of the various diamond bearing kimberlite pipes. Initial open pit mining was replaced by underground mining as the pits went deeper and the last underground mining eventually ceased in 2005 as a result of economic reasons. The mining of these primary resources created a number of tailing resources (previously called dumps) that are scattered all over Kimberley. These resources still contain diamonds as a result of historical inefficient processing technologies. However, each of these resources have unique geological and metallurgical properties because of the differences in the original kimberlites and the different treatment technologies that were used when the tailing resources were deposited. The tailing resources are mined by a fleet of earthmoving equipment and delivered to one of Kimberley Mines‟ five treatment plants for diamond recovery. Each plant is different in terms of treatment capacity, technologies utilised and overall efficiencies. These differences, combined with the geo-metallurgical properties of each individual resource and the hauling distance from the resource to the plant, determine if the specific resource can be treated profitably through that plant. With this array of resources available, Kimberley Mines is in the fortunate position of having flexibility to maximise the Net Present Value (NPV) of the operation. Unfortunately this flexibility also increases the complexity of finding the optimal solution. Mine Planning is currently conducted with a spreadsheet based input – output model but this model is not able to match the resource properties with the plant parameters and this method results in a plan that may be neither practical nor optimal. The model is also very time-consuming and scenario analyses are therefore very limited. This report presents the development of a mixed integer linear programming model to assist with the development of a practical, optimal mine plan. The model is roughly based on a generic model that addresses facility location and processing plant problems as developed by Barbaro & Ramani (1986); however, the model presented in this report includes a more detailed mass balance within the plant and models how the metallurgical properties of the resources affect the mass balances and overall throughputs. The results of the project indicate that the current spreadsheet based Mine Plan is not practical since it violates two mass balance constraints within the Combined Treatment Plant. The newly developed system is used to generate a mining schedule that does not violate any constraints while still delivering the same net present value and overall throughput. The system also illustrates that the plant can improve the annual throughput by 18.2% by changing one screen size. The system also illustrates that only two of the five plants add value and supports management‟s recent decision that the production through the other plants should cease. In summary: The newly developed system generates an optimal, practical mine plan in less than a tenth of the time required for the old spreadsheet based model. The new system can also do various what-if scenarios which the previous model could not answer.

AFRIKAANSE OPSOMMING: Die ontginning van diamante in Kimberley het in die vroeë 1870s begin na die ontdekking van verskeie diamanthoudende kimberlietpype. Aanvanklike oopgroefmynbou is later vervang deur ondergrondse metodes soos die myne dieper geword het. Die laaste ondergrondse myne het in 2005 produksie gestaak as gevolg van ekonomiese redes. Die ontginning van hierdie primêre reserwes het gelei tot die ontstaan van verskeie uitskotreserwes (voorheen genoem afvalhope) verspreid oor Kimberley. Hierdie reserwes bevat steeds diamante as gevolg van die historiese oneffektiewe aanlegte. Die reserwes het elkeen egter verskillende geologiese en metallurgiese eienskappe as gevolg van die verskille in die oorspronklike kimberliete en die verskillende herwinningstegnologieë wat in gebruik was toe die uitskotreserwes gegenereer is. Die uitskotreserwes word gemyn deur ‟n vloot grondverskuiwingstoerusting en word afgelewer by een van Kimberley Myne se vyf aanlegte vir diamantherwinning. Elke aanleg is uniek in terme van kapasiteit, tegnologieë in gebruik en algehele effektiwiteit. Hierdie verskille, tesame met die geo-metallurgiese eienskappe van elke reserwe en die vervoerafstand van die reserwe na die aanleg bepaal of die spesifieke reserwe winsgewend deur daardie aanleg geprosesseer kan word. Bogenoemde verskeidenheid van hulpbronne gee aan Kimberley Myne die voordeel van buigsaamheid om die Netto Huidige Waarde (NHW) van die myn te maksimeer. Hierdie buigsaamheid vergroot egter die kompleksiteit om die optimale antwoord te vind. Beplanning word tans met ‟n sigblad gebasseerde invoer – uitvoer model gedoen, maar hierdie model is nie in staat om die reserwes se eienskappe by die aanlegte se parameters aan te pas nie. Hierdie metode genereer dus ‟n plan wat moontlik nie prakties of optimaal is nie. Die model is ook tydrowend en scenario analises is dus baie beperk. Hierdie verslag beskryf die ontwikkeling van ‟n gemengde heeltallige lineêre programmeringsmodel vir die ontwikkeling van ‟n praktiese, optimale produksieplan. Die model is losweg gegrond op ‟n generiese model van Barbaro & Ramani (1986) wat fasiliteit posisionering en aanlegprobleme aanspreek. Die model in hierdie verslag bevat egter ‟n meer gedetailleerde massabalans binne die aanleg en modelleer hoe die metallurgiese eienskappe van die reserwes die massabalanse en die algehele kapasiteite beïnvloed. Die resultate van die projek dui daarop dat die huidige sigblad produksieplan nie prakties is nie aangesien dit twee massabalansbeperkings in die Combined Treatment Plant oorskry. Die nuwe stelsel is gebruik om ‟n produksieskedule te genereer wat nie enige beperkings verbreek nie en wat terselfdertyd die aanvanklike netto huidige waarde en totale produksie ewenaar. Die stelsel wys ook dat die jaarlikse kapasiteit met 18.2% verhoog kan word deur een van die sif-groottes te verander. Die stelsel demonstreer ook dat slegs twee van die vyf aanlegte waarde toevoeg en ondersteun bestuur se onlangse besluit dat die produksie deur die ander aanlegte gestaak behoort te word. Ter samevatting: Die nuwe stelsel genereer ‟n optimale, praktiese produksieplan in minder as ‟n tiende van die tyd benodig vir die ou sigblad model. Die nuwe stelsel kan ook verskeie “Wat sal gebeur indien…” scenario‟s hanteer wat die vorige model nie kon beantwoord nie.

Please refer to this item in SUNScholar by using the following persistent URL: http://hdl.handle.net/10019.1/50673
This item appears in the following collections: