Sugarcane transportation process modeling by time series approach

Abstract

Sugarcane is one of the severely perishable crops that is used as raw material for white sugar production. Sucrose content of the sugarcane which is of high commercial value decreases in quality due to pre-harvest burning, high ambient temperature, kill-to-mill delays as well as microbial contaminations. Delays in sugarcane transportation are the most important risks which can affect the quality and quantity of the product. Delay in milling of the harvested sugarcane is caused by various reasons in agro-industry units including factory downtime, breakdowns of tractors in the waiting line at factory, tractor accident in factory yard and staff shift changes creating long queues. In order to reduce delays, the present study attempted to forecast arrival and service level of tractor drawn carts which transfer burned or cut canes from farm to mill. The univariate ARMA models were applied to forecast arrival and service level. The RMSE and MAPE were also used to evaluate precision of our forecast. The results of models demonstrated that ARMA(4,3) and ARMA(4,2) models are suitable for arrival and service level of tractor drawn carts, respectively. The predicted values trend of arrival, and service level truly reflected the actual values of arrival and service level as well as queue system tendency. The values of MSE, RMSE and MAPE that indicate accuracy of the forecasted carts arrival and service level were relatively low. The estimated models can be used to forecast values of arrival and service levels of tractor drawn carts for subsequent hours during harvest season.