Inventory policies for demands with trend and randomness. Lubricants marketing case

Abstract

At present, inventories are considered as waste, because they do not generate added value, and so they must be reduced or eliminated. However, this reduction in inventories should not lead to a decrease in the level of customer service or generation deficit. Considering this, this article proposes inventory policies that seek to maintain a previously defined level of customer service with the lowest cost of inventory management. The work is applied in a lubricant trading company whose demand shows a growing trend and is subject to randomization. The forecast model that best fits the demand behavior is defined, and it is statistically verified that the error fits a normal distribution with an average equal to zero. Policy definitions are based on the combination of heuristic techniques for inhomogeneous demands, such as Silver Meal and Wagner-Whitin, to define order periods and models for stochastic demands, such as periodic review models, for multi-product inventory systems, in which, by means of the probability density function of the forecast errors, the security stock of each period is defined. Applying the previously mentioned techniques, it is possible to obtain a policy in which the time between periods is variable as well as the quantity, which could not have been obtained through the use of periodic review techniques. This demonstrates that the definition of inventory policies in real systems requires the integration and adaptation of different inventory models to be efficient.