REVISTA DYNA MANAGEMENT

Key words:

Distribution planning, Bayesian Network, Supply Chain, Monte Carlo Simulation, Mixed-integer programming model, Uncertainty Quantification, Planificación de la distribución, Red Bayesiana, Cadena de Suministro, Simulación de Monte Carlo, Modelo de programación lineal de enteros mixtos, Cuantificación de incertidumbre., Quantification of uncertainty, distribution planning, Bayesian Networks, supply chain management, logistics

Article type:

ARTICULO DE INVESTIGACION / RESEARCH ARTICLE

Section:

RESEARCH ARTICLES

ABSTRACT:
The Effective planning in the distribution of product or consumer goods is necessary to ensure that a supply system can meet the demand in a reliable and economical way; In addition, today competition encourages manufacturers and retailers of consumer goods to differentiate themselves through the specialization of the supply of goods and services that are adapted to one or more subgroups or market segments. However, at the time when distribution decisions are made, many elements of the system are uncertain, such as the level and frequency of demand, the type and location of consumption centers, are an example of this uncertainty; this is due to an incomplete knowledge that arises from the lack of data, measurement errors, lack of resolution, biased sampling and/or natural randomness. At present, decision-making explaining such uncertainties is generally performed considering a small set of plausible scenarios, and the limited coverage resulting from the parameter space limits the confidence of the resulting decision with about the real world. This document presents a methodology that uses statistical emulators to quantify the uncertainty in the results of the mixed-integer linear programming model (MILP), to control the uncertainty in the decision process arising from the set of finite-size product distribution scenarios. Considering the integration of production, inventory and distribution decisions in a supply chain composed of several production facilities that supply several distribution centers that, in turn, provide to retailers located in the same region. The result is to find the optimum levels of production and distribution between a set of factories, warehouses, and points of sale/consumption under uncertainty, considering a Bayesian network that allows us to analyze the relationship between inputs and outputs to identify critical uncertain entries.

Keywords: Distribution planning, Bayesian Network, Supply Chain, Monte Carlo Simulation, Mixed-integer programming model, Uncertainty Quantification