Aplicación de series de tiempo en la realización de pronósticos de producción

Abstract

Production forecasting is a daily activity in Petroleum Engineering, developed by commercial software based in Decline Curve Analysis and Type Curves. These models have failures in two aspects: the first one, models are conditioned to wells producing under pseudosteady state; second one, the production data is fitted to a tendential line, that is extrapolated in the time for get the forecasting. In this investigation, as an alternative to those models, a time series application is proposed, because time series include tendencial, cyclic and stational components of production data. The error between the actual data was compared to the forecasts obtained by conventional methods and results of the time series model. This application allowed to obtain a better history matching of data, evidence that other trends may be in decline (cubic, for example) and increase the accuracy of forecasts generated.

Keywords

• Production
• Forecasting
• Time Series
• ARIMA Model.