Statistical Methods for Modelling Petroleum Products Consumption in Iraq

Abstract

In this research there are models developed to consumption for three major petroleum products; Fuel oil, Gas oil and Gasoline for five years (2016-2020) based on two statistical methods; Autoregressive Integrated Moving Average (ARIMA) and Multi Linear Regression (MLR). The data collected for actual demand and consumption from 2005-2015 years, from different sources are used to estimate petroleum products consumption for Baghdad governorate (as a case study).These predicted models are affected by factors as population, urbanization rate, number of vehicles and electricity sector. The generated results by (ARIMA) are suitable to forecasting and more accurate, since it depends on Mean Absolute Percentage Error (MAPE) to determine forecasting accuracy the results showed that the error was equal to 8.1%, 15.93% and 10.57% for Gasoline, Gas oil and Fuel oil respectively. SPSS program version (23) is used to reveal these results, showing an increase in consumption of Fuel Oil and the stability of consumption to Gasoline and Fuel oil, in the same level of consumption in the past years resulting an increase in the gap between demand and produced quantity when compared to Doura refinery production quantity of these fuels. While the produced quantity of Fuel oil is greater than the predicted consumption required. These results are valuable to decision makers to select between different alternatives as increase in production of light products (Gasoline, Gas oil) and reduce in Fuel oil production or importing these fuels, or the decision of implementing new refinery.