Thesis (MBA)--Stellenbosch University, 2002. / Some digitised pages may appear illegible due to the condition of the original hard copy / ENGLISH ABSTRACT: The Fruit Unit of Tiger Brands is annually responsible for the canning of 75 000 tons of fruit of which apricots constitutes approximately twenty percent. The canning of apricots is subject to a few unique challenges with regard to production planning. The challenges are the unpredictable fruit sizes, unpredictable fruit quality, unpredictable fruit degradation in cold storage, unknown starting date for production and the uncompromisable end of production before Christmas day. These uncertainties led to the fact that no formal production planning for apricots had been done in the previous years of operation. The primary aim of this study is to reduce the risks with regard to production, apricot size and apricot quality by effectively forecasting the expected nett result thereof. This will quantify the resultant products available to marketing personnel and give a measure to evaluate and control production performance. The problem is addressed by organising and presenting historical data such that forecasts of future outcomes would become possible. Clear trends are present on throughput and fruit degradation over time, making forecasting of these two uncertainties quite simple. The forecasting of fruit size is however more problematic, especially because of the lack of sufficient data. It was proposed that the four main fruit size categories be described by using probability distributions fitted over the actual data of the last four years. The large variation on these distributions, probably caused by the lack of sufficient data, rendered this method unsuitable. It was decided that the best estimate of the percentage of each fruit size to be expected, was the average of the derived distributions. The model constructed of the above forecasts is suitable for the estimating of the quantities of specific products that would be available as output from the production process. The model is however not suitable for the evaluation and control of production processes. It is proposed that evaluation and control of production be achieved by using control charts derived from the same historical data. The production control charts are constructed from actual, cumulative production output from the last four years. Linear regression was done on this data to establish a trend line with two sigma limits plotted on the same chart. These charts could be used effectively to monitor daily production output to establish if the commitment towards marketing would be achieved. The lack of data for analysis puts a question mark on the statistical significance of the model. The model is viewed to be a first step in the elimination of uncertainties of raw materials and production variances by making use of historical performance data. The model and control charts will become more and more statistically significant if future actual performance data is incorporated. The model could also be drastically improved if detailed agricultural models for the prediction of apricot size and quality, based on climatic and soil conditions during the growing period were available. The development of such models could be the subject of future studies. / AFRIKAANSE OPSOMMING: Die Vrugte eenheid van Tiger Brands is jaarliks verantwoordelik vir die inmaak van naastenby 75 000 ton vrugte. Appelkose maak twintig persent van hierdie volume uit. Die inmaak van appelkose is onderhewig aan unieke uitdagings ten opsigte van produksiebeplanning. Die uniekheid is gesetel in die onvoorspelbaarheid van vruggrootte, die jaarlikse verskille in vrugdegradering tydens opberging, die onvoorspelbaarheid van die begindatum van produksie en die absolute vereiste om voor Kersdag produksie te voltooi. Hierdie onvoorspelbaarhede het tot gevolg dat daar tot op hede geen formele produksiebeplanning vir appelkose gedoen is nie. Die primêre doelwit van die studie is om die risiko's ten opsigte van produksie, appelkoosgehalte en appelkoosgrootte te verminder, deur die impak daarvan vooruit te skat. Sodoende sal die produkte wat vir bemarking beskikbaar gestel behoort te word ook beter bekend en beheerbaar wees. Die probleem word aangespreek deur geskiedkundige data sodanig te verwerk en te organiseer dat vooruitskattings daarmee gedoen kan word. Baie duidelike tendense ten opsigte van deurset en vrugdegradering oor tyd is deur middel van die geskiedkundige data waarneembaar, wat vooruitskatting van die twee faktore redelik vergemaklik. Die vooruitskatting van vruggrootte bly egter problematies, veral as gevolg van die gebrekkige geskiedkundige data wat beskikbaar is. Daar is gepoog om die vier gespesifiseerde vruggrootte kategorieë deur middel van waarskynlikheidsverdelings te beskryf, maar die gebrekkige data en groot variasie van die data en verdelings maak die tegniek ongeskik. Gevolglik is daar besluit om slegs die gemiddeld van hierdie verdelings as 'n beste raming te gebruik van die verwagte hoeveelheid van elke vruggrootte. Die model wat uit bogenoemde manipulering van data afgelei word, word gereken geskik te wees vir die raming van hoeveelhede van spesifieke produkte wat vir verkope beskikbaar sal wees. Die model is egter onvoldoende vir produksiebeheer en evaluasie. Produksie evaluasie sal egter met behulp van produksiebeheer kaarte wat ook van geskiedkundige prestasie afgelei is, gedoen word. Die produksiebeheer kaarte is kumulatiewe werklike fabrieksprestasie waardeur 'n regressielyn gepas is, met twee sigma afwykingslyne weerskante van die regressielyn geplot. Hierdie kaarte sal gebruik word om daaglikse produksie prestasie te plot en te interpreteer of die kommitment teenoor bemarkingspersoneel steeds haalbaar sou wees. Die model en die produksiebeheer kaarte is afgelei van slegs vier vorige seisoene se produksiedata. Die tekort aan relevante data plaas die betekenisvolheid van die afleidings dus onder verdenking. Die model en kaarte word beskou as 'n eerste poging om die risiko van grondstof- en produksievariasies te verminder deur die vooruitskatting van die uitsette met behulp van geskiedkundige prestasie. Die model en kaarte sal met die inkorporering van toekomstige seisoenale data meer statisties beduidend word. Die model sou ook verbeter kon word deur detail landboukundige modelle wat appelkoos gehalte en grootte verbind met klimatologiese en grondkundige kondisies tydens die verbouingsproses van die vrugte. Hierdie verbetering word egter voorgestel vir 'n verdere studie onderwerp.