Stochastic programming applied to olive husk milling process

Abstract

Olive husk is a by-product obtained from the industrial process for olive oil production containing oil, olive skin and stones. This olive husk is usually carried from presses to an Olive Waste Management Center where a mechanical milling extraction process is performed to obtain Olive Pomace Oil. The remaining oil in the husk is chemically extracted and finally, the residual husk is used as fuel for a cogeneration process to produce electricity. The olive Pomace Oil from mechanical milling has higher quality and price than the one extracted chemically. As the seasonal olive collection campaign is from October to March, husk usually is stored and processed along the year. Due to the own husk chemical degradation and the constrained milling capacity, not all received husk is milled mechanically but always chemically. Different pools are used to store and classify husk qualities. There are pools for short, medium and long term horizons. Short term pools have low storage capacity and contains high quality husk which is shortly milled mechanically. Medium term pools have higher storage capacity containing medium quality husk. These mid term pools provide husk to be milled mechanically when the collection rate decreases, or the husk quality coming from presses is poor. Long term pools have also high storage capacity and store husk with poor quality to be processed only chemically. The stochastic model is focused on optimal pouring decisions into different types of pools to improve the efficiency of the husk mechanical milling. Several probabilistic scenarios of weekly husk pouring from presses are included in the stochastic programming model. Blending husk with different qualities brings nonlinear constraints into the computation of milling efficiency. To cope with these nonlinearities, an accurate approximation has been developed. A real case study is analyzed for different weeks along the collection campaign.